How to Use Open WebUI: A Self-Hosted ChatGPT-Style Interface for Your Business

What It Is

Open WebUI is a self-hosted AI platform that gives your business a ChatGPT-style chat interface running on your own infrastructure. It connects to local models through Ollama as well as OpenAI-compatible APIs, so you can mix local and cloud models in one place. The project describes itself as a home for AI that is extensible, feature-rich, user-friendly, and built to run entirely offline. On GitHub, the repository open-webui/open-webui has more than 155,000 stars, which makes it one of the most widely used open-source AI interfaces available. For a small business, the practical appeal is simple: your team gets a familiar chat experience without sending every conversation to a third-party service by default, and you control where the data lives.

Who It’s For / SMB Use Case

Open WebUI fits small businesses that want a private, shared AI workspace. A marketing agency can give every team member access to the same models with their own logins. A clinic, law office, or consultancy can keep internal questions and documents on its own server. A shop owner can run a small local model for drafting replies and product descriptions without paying per-seat fees for a hosted chatbot. It also works well as a front end for a team that already runs Ollama, or for a business that uses several model providers and wants one consistent interface. If you have no one comfortable with basic server setup, the learning curve is real, but the payoff is a chat tool you fully control.

Installation

There are two main ways to install Open WebUI, both documented in the project README. The pip route needs Python 3.11. The Docker route works on Windows, macOS, and Linux, as long as Docker is installed. The README warns that your Docker command must include a data volume, or you risk losing your database. Pick one route below.

Option 1: Install with pip

1. Make sure Python 3.11 is installed, as the README recommends it to avoid compatibility issues.
2. Install the package:

   pip install open-webui

3. Start the server:

   open-webui serve

4. Open http://localhost:8080 in your browser and complete the on-screen setup.

Option 2: Install with Docker

If Ollama already runs on the same computer, use this command from the README:

docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

If Ollama runs on a different server, set the OLLAMA_BASE_URL variable to that server’s address. For a single container that bundles Open WebUI with Ollama, the README offers the :ollama image, for example for CPU-only machines:

docker run -d -p 3000:8080 -v ollama:/root/.ollama -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:ollama

GPU users can use the :cuda image, but the README notes that this requires the Nvidia CUDA container toolkit on Linux or WSL. After the Docker install, open http://localhost:3000. If the container cannot reach Ollama, check the official troubleshooting docs, which describe the --network=host option.

How to Use

1. Finish the first-run setup. Open the address above and follow the on-screen steps to create your workspace. If you plan to share it with staff, review the user and role settings in the admin panel. For exact admin options, check the official docs at docs.openwebui.com.
2. Connect a model. If you used the bundled :ollama image, Ollama is already included. If you installed Open WebUI alone, point it at your Ollama server or add an OpenAI API key, or connect another OpenAI-compatible provider such as LM Studio, Groq, Mistral, OpenRouter, or vLLM. The README lists these as supported connection targets.
3. Start a chat. Pick a model in the chat screen and ask your question. You can run several models in one conversation to compare answers, a feature the README calls multi-model conversations. Save useful prompts and reuse them across the team.
4. Bring in your own documents. Use the # command in chat to pull a document from your library, or a URL, into the conversation. Open WebUI supports local RAG with hybrid search and reranking, so answers can draw on files you uploaded rather than only on the model’s training data.
5. Build a reusable assistant. Wrap a base model with custom instructions, tools, and knowledge to create a specialised assistant, for example a support-reply drafter or a quoting helper. The README also documents plugins, filters, actions, pipes, tools, and MCP connections for deeper integrations. For step-by-step details on any of these, check the official docs.

Pricing / Limitations

Open WebUI itself is open-source software you can self-host without a licence fee, but running it is not free in practice. You pay for the computer or server it runs on, plus any API usage if you connect paid cloud models. Local models cost no per-message fee, but they need enough CPU, RAM, or GPU power to answer at a usable speed. The README also mentions an Enterprise Plan, so if you need enterprise features or support, check the official docs for current pricing. On licensing, the README states that the project contains code under multiple licences, including the Open WebUI Licence with a requirement to preserve the Open WebUI branding, plus prior contributions under their original licences. Read the LICENSE and LICENSE_HISTORY files before you rebrand or redistribute it. Finally, self-hosting means you own updates, backups, and security, so plan for that ongoing work.

Bottom Line for SMB

Open WebUI is the fastest credible way to give your whole team a private, ChatGPT-style workspace without per-seat chatbot fees. Start with the Docker :ollama image, connect one small local model, and test it on one real workflow, such as drafting replies or summarising documents. If the team uses it daily, expand from there.

FAQ

Does Open WebUI work without the internet?

Yes, the project says it is built to run entirely offline when paired with local models through Ollama. You can also set the HF_HUB_OFFLINE environment variable to 1 to stop model download attempts in offline environments.

Do I need Ollama to use Open WebUI?

No. Open WebUI works with Ollama for local models, but it also connects to OpenAI and other OpenAI-compatible APIs. If you only use the OpenAI API, the README shows a Docker command that passes your API key with OPENAI_API_KEY.

Can my team share one Open WebUI install?

Yes. The README documents user accounts, groups, and granular role-based access control, so an admin can give each person exactly the access they need. For the exact settings, check the official docs.

Related reading: AI Agents Explained for Small Business and ChatGPT vs Claude vs Gemini for Small Business.